AIG - American International Group

[vinod1]

The exact calculations required to complete this are outlined in the book, "Financial Instruments and Institutions: Accounting and Disclosure Rules" by Stephen Ryan, 2nd Edition. Specifically Chapter 15, "Property Casualty Insurer's Loss Reserve Disclosures". He walks you through the transformations needed to get the tables I have listed. I do not think I would be able to explain, as it is a bit difficult to do so on a message board. I would see if I can scan the 10 odd pages he devotes to this calculation.

 

Let me explain a little bit what I was trying to get at. I see two problems in analyzing the reserve tables for any insurance companies

 

1. Luck/Randomness - A company could have reserved appropriately but bad luck could have caused higher losses or vice versa. I do not think we can really ever be 100% of reserving on long tail exposure due to this. A long time period would increase our confidence but we can never be certain. There is nothing we can do about this other than being less certain in our confidence on any reserve data from a long tailed P&C company.

 

2. A P&C company establishes/modifies reserves each year both (a) for policies written this year and (b) for any changes in estimates of losses for past policies. In any given year management can play with estimates for either (a) or (b). As time passes reality catches up and management has less and less discretion in managing reserves for (b). The problem I am trying to answer is if management has conservatively estimated the reserves for (a) i.e. for policies written for a particular year. So you are looking at the reserve number management has estimated for a given year and then see how it has evolved subsequently just for that specific year.

 

What you ideally want to do is to look at management estimates for policies written in a particular year and see how the reserves allocated for that particular year have developed. So you can sort of have independent "exams" so to speak of management estimates for reserves. This also eliminates the problem of reserve deficiency that was experienced in one year from contaminating other years.

 

The conventional loss reserve data presented in 10-K's do not lay this out clearly, although the data to perform the analysis is all there. When you look at the reserve tables laid out in a 10-K, the deficiency is a jumble of past and present reserve changes and it provides very limited help without further analysis.

 

For example, a company could have been reserving very conservatively for 9 of the past 10 years but one year it could have had a very bad experience and analyzing as above would help answer this kind of question. We can say management has been conservative in 9 out of the 10 years and has reserved sufficiently in all these 9 years. We still have Luck/Randomness problem so we would never be sure.

 

With the above approach you can see trends in reserving by individual years.

 

Vinod

 

[PlanMaestro]

Simply put, this company is too valuable to ignore. It's too valuable to ignore even where we stay today with a market cap north of $40 billion, with a book capital of $80 billion-plus, which by the time we've removed the reserve on the DTA, by the time we have taken the DAC write off in EITF 09-G, we'll be roughly speaking $100 billion of book capital in the not-too-distant future.

 

Peter Hancock. CEO Chartis

 

What is EITF 9-G?

http://www.soa.org/library/newsletters/financial-reporter/2010/september/frn-2010-iss82-bott.pdf

[PlanMaestro]

My understanding is that actuaries would say that triangles are not enough and that the 2008 numbers are a mess considering all the things going on at the time.

 

http://www.chartisinsurance.com/aigeurope/internet/it/files/statement%20a%20commento%20dell'analisi%20di%20Bernstein%20Research%20sulle%20riserve%20AIG%20-%20Chartis_tcm762-229069.pdf

[vinod1]

My understanding is that actuaries would say that triangles are not enough and that the 2008 numbers are a mess considering all the things going on at the time.

 

http://www.chartisinsurance.com/aigeurope/internet/it/files/statement%20a%20commento%20dell'analisi%20di%20Bernstein%20Research%20sulle%20riserve%20AIG%20-%20Chartis_tcm762-229069.pdf

 

I fully agree with the sentiment. Triangles are not enough, I am just trying to extract as much of the info as possible from the data that is presented. Just like we do an analysis of the composition of the loan portfolio to determine if charge offs are reasonable and not just rely on the overall charge offs for a bank.

 

Vinod

[PlanMaestro]

I fully agree with the sentiment. Triangles are not enough, I am just trying to extract as much of the info as possible from the data that is presented. Just like we do an analysis of the composition of the loan portfolio to determine if charge offs are reasonable and not just rely on the overall charge offs for a bank.

 

On the same page Vinod, all this information helps to get a better understanding of what is going on but it is not perfect. A couple of more datapoints:

 

1. GAO March 2009: While we have found no evidence that federal assistance has been provided directly to AIG’s property/casualty insurers, as has been the case for AIG life insurers, AIG’s insurance companies have likely received some indirect benefit to the extent that the property/casualty insurers would have been adversely affected by a credit downgrade or failure of the AIG parent.

More details: http://www.gao.gov/new.items/d09490t.pdf

 

2. Pennsylvania and New York DOI investigation Dec 2010: The department concluded that the Chartis businesses' loss reserves "were in the range of reasonable estimates" after the company raised its projections of losses in 2009. It also said the units' pricing of insurance policies was in line with competitors.

More  details: http://online.wsj.com/article/SB10001424052970203525404576050572352208308.html

http://www.programbusiness.com/news/Chartis-Taking-Steps-to-Improve-Disclosures-After-Regulator-Finds-Deficiencies

http://www.riskandinsurance.com/story.jsp?storyId=533328604

 

3. Several reserves increase and accounting improvements:

http://www.insuranceheadlines.com/pdf/7117.html

 

4. David Merkel: he did an extraordinary job going through the statuatory books and found many instances of hidden leverage beyond FP and Securities Lending. And despite that he was very criticial of AIG (He thought common a 0, Maiden Lane 0 recovery), he thought P&C was mostly OK.

 

To what degree were AIG’s operating subsidiaries sound? Answer: aside from the mortgage insurers, the P&C subsidiaries were basically sound, though with some issues such as capital stacking, affiliated assets, etc., as mentioned above. The non-mortgage P&C subsidiaries didn’t have a great 2008, but they would have survived as standalone entities. The life and mortgage subsidiaries are another matter. Without the help of the US Government, many of them would have failed. Even now, given the levels of affiliated assets, capital stacking, deferred tax assets, etc., they are not in great shape now should there be another surprise. Profitability is likely to be lower in the future than in the banner years of the middle of the 2000s decade.

More: http://seekingalpha.com/article/134260-to-what-degree-were-aig-s-operating-insurance-subsidiaries-sound-part-1

 

5. Behmosche had to go out of his way to not held the US government responsible for AIG's reserves in the re-IPO, when he had all this scrutinity.

More: http://www.bloomberg.com/news/2011-05-24/aig-says-statements-to-potential-investors-need-clarification-.html

 

[vinod1]

In testimony to COP:

 

Mr. SILVERS. Mr. Benmosche, I’m sort of interested in the contradiction or the contrast between your testimony, Mr. Gallant’s testimony, and the testimony of Mr. Clark from S&P. Can you (a) explain to me your understanding of the difference between your estimation of the company’s earning power going forward, after your asset sales, and Mr. Gallant’s? And can you (b) explain to me if your general characterization of your company’s financial position is consistent with S&P’s view that absent government support you’re Double B?

 

Mr. BENMOSCHE. I can’t comment on Mr. Gallant, you’ll have to get him to figure it out. I know what I’m running, I know the company I’m running, and I have confidence in this company, and I know what I’m talking about. So you’ll have to see whether he understands the company as well as I do.

 

Mr. SILVERS. Mr. Benmosche, that’s not an acceptable answer.

Mr. BENMOSCHE. Okay.

 

Mr. SILVERS. You know we represent your majority stockholder,

or at least we kind of do. We are trying to look out for your majority

stockholder. I am frankly frightened by what Mr. Gallant said

on behalf of the American public. And I would like you to explain

specifically with reference to numbers why he’s wrong.

 

Mr. BENMOSCHE. I have not looked at his report. I’d be glad to have a team of people study it and do a side-by-side. We did that when we had a report that said that we had an $11 billion hole in our reserves. It was written by Bernstein, and in fact, you found out we did not have an $11 billion hole. We actually went through that report, showed them why they were wrong, and they still went forward with it. So I’m happy to do that for him as well.

 

Plan,

 

You seem to have a knack for finding obscure documents. I am not sure if the above report by AIG on the reserves is public. It would be very interesting to get our hands on that report.

 

Vinod

 

[PlanMaestro]

I am not sure if the above report by AIG on the reserves is public. It would be very interesting to get our hands on that report.

 

Will try, but the Towers Watson report was not public and not even AIG knows the methodology. Do you have a link to the testimony?

 

Found Silvers vs Benmosche: http://www.gpo.gov/fdsys/pkg/CHRG-111shrg63515/html/CHRG-111shrg63515.htm

[vinod1]

I am not sure if the above report by AIG on the reserves is public. It would be very interesting to get our hands on that report.

 

Will try, but the Towers Watson report was not public and not even AIG knows the methodology. Do you have a link to the testimony?

 

Found Silvers vs Benmosche: http://www.gpo.gov/fdsys/pkg/CHRG-111shrg63515/html/CHRG-111shrg63515.htm

 

Testimony is worth reading. I must have read it a couple of times. When do you get an opportunity to get a CEO to go in front of congress and testify under oath as to the normalized earnings power of the company?

 

Link to testimony http://frwebgate.access.gpo.gov/cgi-bin/getdoc.cgi?dbname=111_senate_hearings&docid=f:63515.pdf

 

Source page with other links http://www.gpoaccess.gov/congress/joint/panel/index.html

 

Main site for COP AIG material: http://cybercemetery.unt.edu/archive/cop/20110401232818/http://cop.senate.gov/reports/library/report-061010-cop.cfm

 

Vinod

[pdoak]

Vinod

 

I am very interested in the table of loss reserves that you posted.  I appreciate that it may be difficult to explain but would you mind posting the spreadsheet behind the pdf so that it is possible to see how the numbers are calculated and tie them back to the 10K.

 

Many thanks

[PlanMaestro]

Testimony is worth reading. I must have read it a couple of times. When do you get an opportunity to get a CEO to go in front of congress and testify under oath as to the normalized earnings power of the company?

 

And Benmosche was probably low-balling AIG's earnings power but nobody believed him the number!

 

And so I believe that you're looking at a company that once we sell off the companies that we've talked about, or assets that we've talked about, I still think we're talking about a  company that could earn, in 2011, without extraordinary charges and goodwill charges, and all these other things, I believe we  have a company that can earn between $6 billion and $8 billion after taxes.

 

This other part is also interesting. AIG has gone very far in de-risking the balance sheet. This was the situation in September 2008 according to Jim Millstein, Chief Restructuring Officer, Treasury:

 

All of the contracts at AIGFP are guaranteed by the parent. The parent has a $100 billion dollar balance sheet of its own. On September 8th of 2008, with $15 billion dollars of  commercial paper, we all know what happened to Lehman Brothers,  to the commercial paper markets after Lehman Brothers filed and defaulted on $5 billion dollars of commercial paper.

 

Fifteen billion dollars of commercial paper at the parent  company. Eighty billion dollars of repo. Again, the repo  markets went into seizure after the Lehman Brothers filing. And  a much smaller amount of repo. Two trillion dollars of notional derivatives, $400 billion of credit derivatives, concentrated  very much in the real estate part of the market.

 

Had AIGFP defaulted on the collateral posting requirements  that it had on September 16, every counterparty, 44,000 trades could have terminated their trades, declared cross default--

 

What is the situation now? AIGFP mostly gone page 43. Parent company current balance sheet page 44:

http://www.aigcorporate.com/investors/2011_November/Q32011_FinancialSupplement.pdf

[CONeal]

Have been looking through the testimony and would like some clarification for those that have more experience in the insurance industry.

 

S&P states that right now AIG has a rating of BB, due to the government backstop is A-.  What makes you confident that once the backstop s gone that AIG's rating will be raised to A-?

 

Also, what should the rating be in order to attract people for the high net worth life insurance policies?

[vinod1]

Vinod

 

I am very interested in the table of loss reserves that you posted.  I appreciate that it may be difficult to explain but would you mind posting the spreadsheet behind the pdf so that it is possible to see how the numbers are calculated and tie them back to the 10K.

 

Many thanks

 

Pdoak,

 

I did the calculations manually (that is why I did not calculate the complete triangle for the 2nd table), I just used excel as it is easier to put tables there. There are not formulas in it. So what you see in pdf is what you have in the excel.

 

Vinod

[PlanMaestro]

A reader sent me the transcript of the Peter Hancock presentations in December. Very interesting.

 

http://dl.dropbox.com/u/32606407/AIG_US_2012-01-18_PRESENTATION.pdf

[PlanMaestro]

S&P states that right now AIG has a rating of BB, due to the government backstop is A-.  What makes you confident that once the backstop s gone that AIG's rating will be raised to A-?

 

Killer presentation today by Benmosche. Check first 6 pages on financial condition including rating agencies ... ratings.

 

http://www.aigcorporate.com/investors/BOFAMerrillLynch_InvestorPresentation.pdf

[Uccmal]

Mfc and slf have released results last week and just now.  Both are getting tortured by the low interest rates. 

 

Aig reports next week.  I expect the life results will look similar to the above.  Maybe a little premature to back up the truck.

[PlanMaestro]

Mfc and slf have released results last week and just now.  Both are getting tortured by the low interest rates. 

 

Aig reports next week.  I expect the life results will look similar to the above.  Maybe a little premature to back up the truck.

 

SunAmerica is a small part of the value thesis. They are raising rates in Chartis to achieve CORs below 95% so they do not have to rely on interest income. Plus AIA, ILFC, United Guaranty, ML II and III, and the deferred tax allowance.

 

They will probably get hit by Thailand though in this report.

[vinod1]

S&P states that right now AIG has a rating of BB, due to the government backstop is A-.  What makes you confident that once the backstop s gone that AIG's rating will be raised to A-?

 

Killer presentation today by Benmosche. Check first 6 pages on financial condition including rating agencies ... ratings.

 

http://www.aigcorporate.com/investors/BOFAMerrillLynch_InvestorPresentation.pdf

 

1. Very nicely laid out. I wish they just complete the circle and lay out the whole asset value and earning power in one slide.

 

2. Me thinks, many of the investors who would be investing in AIG would not invest this time around purely for career risk reasons. I would be very surprised to see for example Chris Davis invest in this again. He got burnt the first time and if he gets burnt the second time as well, everyone would be asking "Did you not learn the lesson the first time?". So I would expect this dynamic to play out with institutional investors and the stock may lag for a while (this is the market timer in me talking).

 

3. The Net residual value remaining on slide 9, it is already reflected on the books so it is already included in the book value. All these assets are being carried at fair value and marked to market so it does not change the thesis much. It just shows that these assets could be monetized and cash deployed elsewhere.

 

Vinod

 

 

 

 

[PlanMaestro]

OK, Let's throw some numbers to the ring to get the conversation started

 

1. AIA 5B: 1/3 share marked to market is 13.5B, net of 8.4B of government preferred interest.

2. ILFC 7B: waiting for spinoff that would give us a market value. Speculation 7B-13B, let's take the low end.

3. Maiden Lane II + III 7B: it will probably be more (w/interests) but let's leave it at that because it's complex.

4. Met escrow 2B

5. Parent 0B: I will consider it a wash despite Parent having $17B in ST investments and $3B in equity

6. United Guaranty and others 0B: to be conservative.

 

So just in non-core assets AIG has more than $21 billion ($11 per share)

 

7. Chartis 70B: used to earn 7-9B  after tax and COR in low 90s. But to be conservative let's use low estimate and multiple.

8. SunAmerica 16B: was earning 2B+, Benmosche thinks it can earn 4B, and over the last year has generated 1.2B just in dividends to the Parent. However, people are worried because of the low rate environment so let's not pay more than AIG paid in 1998 ... 16B.

9. DTA 26B

 

So the total is around 130B ($72 per share), and 105B ($55 per share) without the DTA. The government has no recourse beyond his large common share ownership. There must be something wrong, right?

 

 

[prunes]

Maybe a naive question: How sustainable are Chartis's earnings in the current interest rate environment? The company probably owns a lot of bonds purchased when yields were higher. As these bonds mature the proceeds have to be reinvested at a lower yield.

[pdoak]

I was very interested in the reserve table that Vinod posted and have created an Excel spreadsheet calculating all the figures from the annual report reserve tables.  The spreadsheet can be used to look at other insurance company's reserve deficiencies by updating the loss reserve table for the desired company.

AIG_Loss_Reserves_2010.xlsx

[vinod1]

I was very interested in the reserve table that Vinod posted and have created an Excel spreadsheet calculating all the figures from the annual report reserve tables.  The spreadsheet can be used to look at other insurance company's reserve deficiencies by updating the loss reserve table for the desired company.

 

Good work!

 

Vinod

[vinod1]

OK, Let's throw some numbers to the ring to get the conversation started

 

1. AIA 5B: 1/3 share marked to market is 13.5B, net of 8.4B of government preferred interest.

2. ILFC 7B: waiting for spinoff that would give us a market value. Speculation 7B-13B, let's take the low end.

3. Maiden Lane II + III 7B: it will probably be more (w/interests) but let's leave it at that because it's complex.

4. Met escrow 2B

5. Parent 0B: I will consider it a wash despite Parent having $17B in ST investments and $3B in equity

6. United Guaranty and others 0B: to be conservative.

 

So just in non-core assets AIG has more than $21 billion ($11 per share)

 

7. Chartis 70B: used to earn 7-9B  after tax and COR in low 90s. But to be conservative let's use low estimate and multiple.

8. SunAmerica 16B: was earning 2B+, Benmosche thinks it can earn 4B, and over the last year has generated 1.2B just in dividends to the Parent. However, people are worried because of the low rate environment so let's not pay more than AIG paid in 1998 ... 16B.

9. DTA 26B

 

So the total is around 130B ($72 per share), and 105B ($55 per share) without the DTA. The government has no recourse beyond his large common share ownership. There must be something wrong, right?

 

How can we add up assets and earnings power separately? This is the part that is not clear to me. AIG's portion of investment in Maiden lane, I think are allocated and carried by Insurance subs. There are lots of cross holdings between the subs so I am not sure how to cleanly segregate the assets to come up with sum of parts both from an asset and earnings power perspective - at least without going through each individual insurance sub filings.

 

I think Chartis used to earn 7-9 billion when it has been combined with lots of other subs that have now been sold off. I cannot imagine it earning that much with its current assets. It had $48 billion in statutory surplus and to earn 9 billion means a near 20% ROE. CEO is talking about a 10-12% ROE for Chartis in 2015. I am sure he is low balling but still.

 

Vinod

[bmichaud]

OK, Let's throw some numbers to the ring to get the conversation started

 

1. AIA 5B: 1/3 share marked to market is 13.5B, net of 8.4B of government preferred interest.

2. ILFC 7B: waiting for spinoff that would give us a market value. Speculation 7B-13B, let's take the low end.

3. Maiden Lane II + III 7B: it will probably be more (w/interests) but let's leave it at that because it's complex.

4. Met escrow 2B

5. Parent 0B: I will consider it a wash despite Parent having $17B in ST investments and $3B in equity

6. United Guaranty and others 0B: to be conservative.

 

So just in non-core assets AIG has more than $21 billion ($11 per share)

 

7. Chartis 70B: used to earn 7-9B  after tax and COR in low 90s. But to be conservative let's use low estimate and multiple.

8. SunAmerica 16B: was earning 2B+, Benmosche thinks it can earn 4B, and over the last year has generated 1.2B just in dividends to the Parent. However, people are worried because of the low rate environment so let's not pay more than AIG paid in 1998 ... 16B.

9. DTA 26B

 

So the total is around 130B ($72 per share), and 105B ($55 per share) without the DTA. The government has no recourse beyond his large common share ownership. There must be something wrong, right?

 

How can we add up assets and earnings power separately? This is the part that is not clear to me. AIG's portion of investment in Maiden lane, I think are allocated and carried by Insurance subs. There are lots of cross holdings between the subs so I am not sure how to cleanly segregate the assets to come up with sum of parts both from an asset and earnings power perspective - at least without going through each individual insurance sub filings.

 

I think Chartis used to earn 7-9 billion when it has been combined with lots of other subs that have now been sold off. I cannot imagine it earning that much with its current assets. It had $48 billion in statutory surplus and to earn 9 billion means a near 20% ROE. CEO is talking about a 10-12% ROE for Chartis in 2015. I am sure he is low balling but still.

 

Vinod

 

Plan, thanks for posting the presentation. This definitely makes the analysis more of a "2-foot hurdle". I started taking a look at this this morning, so I'm going to jump into the discussion...

 

Vinod - that's exactly what I was wondering. Is part of the $4 to $5 billion (let's call it $4.5B) of insurance company dividends cited on page 6 of the presentation generated from SPV income or is it purely from insurance operations?

 

This is probably a dumb question with regard to insurance company dividends, but would the $4.5B be after-tax once it reaches the parent or is it pre-tax then AIG pays taxes on it at the parent level? If it is pre-tax and AIG pays 35%, then those dividends could be worth around $59 billion assuming a 5% yield (Ke = 8%, G = 3%).

 

Another question for the group. Given how much AIG has cleaned up its balance sheet (I'm stunned at the reduction in derivitives on page 8 ) and Buffett's penchant for snuggling up to the US Government, would there be a possibility Buffett comes in and buys out some if not all of the UST's investment maybe at 80 to 90% of BV? 85% of BV would probably end up being quite a bit better than what the UST can expect to realize on its own, particularly if it considers how long it will take to exit the position.

 

Fascinating situation. Looking forward to more in depth discussion by the board!

[racemize]

I'm still digging and have not kept up with Plan's rate of data consumption.  However, it seems like we can look at this a little bit more easily.  Let's assume they get to 12% ROE and we're buying at 60% of book value--that gives us 20% ROI while we're buying at, presumably, much less than liquidation value. 

 

I guess that falls under two assumptions:

1) that they can get to 12% ROE, which essentially presumes that they can get to operating profitably; and

2) that long tail doesn't take them out.

 

I'm considering a 9% position.

[PlanMaestro]

Transcript of Benmosche's presentation, first time I've seen him playing offense and he was on fire:

 

http://dl.dropbox.com/u/32606407/AIG_US_2012-02-15_PRESENTATION.pdf